Technological-personal factors of university students’ behavioral intention to continue using online services after the pandemic

Abstract

We examine the role of technological-personal factors of university students’ behavioural intention (BI) to continue using online services provided by the Ministry of Education in Oman after the pandemic. In total, 657 valid responses were collected from students who use online ministry services. We found that self-efficacy (SE) to be a determining factor in their intention to continue using the online services, which is affected by technology task fit (TTF), facilitating conditions (FC) and system quality (SQ) which has the highest effect. Our study is one of the few studies that contributed towards extending the theory of planned behaviour (TPB) after the pandemic, by including the previous four technological-personal factors. We believe that our study will help educational administrative departments to consider the main factors, mainly the quality of online systems, when designing their services to ensure that students continue using online services in the future. The study contributes also to the current research of university students’ behaviour towards using online services provided by central government.

Keywords:

• theory of planned behaviour
• technology task fit
• system quality
• e-government
• e-services

1. Introduction

The rapid rise of Information Technology (IT) use has transformed the way governments serve their citizens (Kurfalı et al., 2017), and regulate the process of creating digital information for governmental services (Ndou, 2004) providing many benefits (Kurfalı et al., 2017). However, the core factors that contribute towards expecting and understanding students’ intention to use e-government services have not yet been identified comprehensively (Alarabiat et al., 2021).

Many generic theoretical models have been developed to identify and examine the factors that affect users’ behaviour towards adopting and using specific IT. For example, Technology Acceptance Model (TAM) developed by (Davis, 1989); theory of planned behaviour (Ajzen, 1991); theory of reasoned action (Ajzen & Fishbein, 1980); unified theory of acceptance and use of technology (UTAUT) developed by (Venkatesh et al., 2003); innovation diffusion theory (Rogers, 1962); and TTF model, as alternative to TAM, which measures how users’ perceived fit of IT for tasks can improve job performance as a result of the actual fit/match of IT with respect to the required tasks (Goodhue & Thompson, 1995), and recently Theory of Planned Behavior (TPB) (Alarabiat et al., 2021).

Ozen et al. (2018) concluded that the most frequently utilized model in explaining the acceptance of online government services is TPB which found to be strong in investigating people’s behaviors in relation to specific issues (Ajzen, 1991). Consequently, this allows many authors to compliantly study the effects of further factors on students’ BI, and hereafter why TPB has become a well adopted model in e-government acceptance research (Alarabiat et al., 2021; Ozen et al., 2018). The flexibility of TPB allows researchers to add any harmonizing factors while keeping the essence of its theoretical reliability (Alarabiat et al., 2021; Alraja, 2022), especially that it has shown robust predictive capability in BI across different contexts (Al-Debei et al., 2013).

Despite the critical positive role of SE in influencing perceived behaviour control (PBC) of adopting e-government services (Hung et al., 2006, 2009; Suki & Ramayah, 2010; Susanto & Goodwin, 2011; Zhao & Khan, 2013), and its effect on the attitude (Verkijika & De Wet, 2018), yet none of the previous studies has examined the BI of using e-services by people and how it can be affected by SQ, TTF, and FC. Therefore, this study has extended Ajzen’s (1991) TPB by adding four constructs (SQ, FC, TTF and SE) as a foundation for forecasting students’ intentions to continue using the online government service after the pandemic. Additionally, and to the best of the authors’ knowledge, no studies has investigated the causal effect of both SQ and FC on BI of online government services considering three key personal variables: attitude (ATT), SE and PBC.

In this paper, we developed an integrated model with a robust combination of technological factors (SQ, FC and TTF), and how they affect personal factors (ATT, PBC and SE). Our aim is to predict university students’ BI of online government services after the pandemic. The extended model explores the effect these variables have upon the BI as perceived by students.

2. Literature review

The e-government research has investigated many aspects of new technologies that are used to provide quality services and improve public administration and economic wellbeing (Baudier et al., 2022; Tejedo-Romero et al., 2022; Ullah et al., 2022). Though earlier studies on e-government provide general perceptions on the quality of online services, more research on this topic is still essential (Alruwaie et al., 2020; Wirtz & Piehler, 2015), especially in the context of developing countries such as Oman. Such countries transformed manual government services into electronical versions (Susanto & Aljoza, 2015), however, online government service is still in its infancy, and faces various challenges in such countries (Kurfalı et al., 2017; Mohammed et al., 2017). In Oman, since 2000, the government has heavily invested in e-services, but its adoption has been thwarted by several factors (Al-Hadidi, 2010). Moreover, few studies of e-government services have been conducted in the context of developing countries. Tarhini et al. (2015) investigated the impact of social, organisational and individual factors on university students’ acceptance of electronic systems. Mohammed et al. (2017) emphasised the importance of increasing awareness and the benefits of e-government services among stakeholders in developing countries, and Abbas et al. (2016) measured the influence of social factors upon students’ intention to use electronic learning services.

Our theoretical and practical contributions to this study are established based on TPB complemented by other essential system technological-personal factors as it provides crucial behavioral perspective of intentions for adopting technology especially when exploring the student motivations (Scuotto et al., 2020). TPB claims that an individual’s intention is anticipated by three key factors: PBC, ATT and subjective norms. While BI refers to an individual’s willingness to undertake a certain behavior; ATT defines the individual’s preferred or unpreferred judgement or evaluation of the behavior under investigation; subjective norms relate the perceived social gravity to achieve or not to achieve certain behavior; and finally PBC defines the degree to which an individual can control their engaged behavior (Ajzen, 1991, 2011).

Several scholars have found that FC (Hung et al., 2009; Kurfalı et al., 2017; Venkatesh et al., 2003), TTF (Escobar-Rodriguez & Monge-Lozano, 2012; Lin & Huang, 2008; Wu & Chen, 2017), and SQ (Insani et al., 2018) are the critical technological factors which require more investigation in the field of e-government. Some researchers concluded that SE has a positive effect on ATT (Verkijika & De Wet, 2018); while there is still a need to explore connections between TTF, SQ and FC from one side and the ATT, SE and PBC form another side. On the same trend, many studies also found that SE and PBC both have a positive effect on BI (Bwalya, 2012). However, despite some studies investigating the relationship between TTF and the ease or difficulty of use of specific technology, which is in line with PBC (Escobar-Rodriguez & Monge-Lozano, 2012; Wu & Chen, 2017), we found no single study inspect the direct relationship between TTF and PBC. In addition, previous studies stated that TTF is one of the main antecedents of an individual’s beliefs about a system’s importance, usefulness and returns gained from using any technology (D’Ambra et al., 2013; Goodhue & Thompson, 1995; Tripathi & Jigeesh, 2015). Similarly, TTF is also found to influence system utilization, which are the beliefs about the results of using a certain system, and worker capability of using this system (Tripathi & Jigeesh, 2015). Accordingly, it is expected that TTF can affect SE (Eom & Estelami, 2012).

In previous studies into, neither the BI of adopting online government services in developing countries, nor how it can be affected by crucial system technological factors such as TTF, SQ, and FC have been given enough attention. Additionally, although extensive research has been carried out in this field, no study has investigated the role of the previously mentioned technological factors together on SE, PBC, and ATT. Thus, our study aims to explore how the online government service affects a user’s BI towards using these services in developing countries, by understanding how the main technological system factors (TTF, FC and SQ), can influence internal personal factors (SE, ATT and PBC). Our model (see Figure 1) has extended the TPB model and is empirically validated by conducting a quantitative research study on online services provided to Omani students by the Ministry of Education.

Figure 1. Research model for students behavioral intention to continue using online services after the pandemic.

3. Hypotheses development

3.1. The impact of technological factors on personal factors

3.1.1. System Quality (SQ)

SQ refers to the functionality of the learning system in terms of its features, functions, content speed, interaction and capability (Fathema et al., 2015), currency, accuracy and efficiency (Kim et al., 2008). In government services, SQ refers to the flexible access, interactivity, user interface features, speed and responsiveness of the e-government portal (Cho et al., 2009). The quality of provided services is divided into offline and online service quality. The quality of offline service moderates the relationship between citizen satisfaction and the quality of online service (Hu & Liu, 2022; Wang & Teo, 2020).

SQ was found to positively influences users’ attitudes towards adopting diverse types of technological e-systems (Insani et al., 2018; Shin, 2017), affects the SE of various e-service systems (Alruwaie et al., 2020), and influences PBC (Lee-Partridge & Ho, 2003). Accordingly, our first set of hypotheses are:

H1:

SQ has a significant positive impact on students’ ATT.

H2:

SQ has a significant positive effect on students’ SE.

H3:

SQ has a significant positive impact on students’ PBC.

3.1.2. Facilitation Conditions (FC)

Organisational factors affect an individual’s perceptions of the ease in executing a particular task (Teo, 2010). FC is the degree to which a person trusts the organisational infrastructure when using a particular system (Venkatesh et al., 2003). It determines the accessibility or availability of key resources (technical/online support, internet availability/infrastructure … etc) that help using the government e-services. Previous studies have laid out the significant impact of FC upon ATT (Fathema et al., 2015; Teo, 2010), on users’ SE (Hung et al., 2006, 2009; Lee-Partridge & Ho, 2003; Suki & Ramayah, 2010; Susanto & Goodwin, 2011), and PBC (Abdullah & Toycan, 2017; García Botero et al., 2018; Tarhini et al., 2015; Teo, 2010; Usoro et al., 2014). Few researchers have focused on finding correlations between FC and individual’s SE in the e-government context (Kumi et al., 2012; Zahid & Din, 2019). Consequently, our second set of hypotheses are:

H4:

Students’ ATT is significantly affected by FC.

H5:

Students’ SE of online services is significantly affected by FC.

H6:

Students’ PBC of using online services are significantly affected by the ATT.

3.1.3. Technology Task Fit (TTF)

TTF has been used widely in the field of IS (Lin, 2012; Lin & Huang, 2008; Lu et al., 2014), however, its usage in the e-government environment, in particular from the students’ perspective, is still uncommon (Roth et al., 2022). TTF highlights the benefits and anticipated results from using e-government services (Goodhue & Thompson, 1995), therefore, TTF can positively affect the ATT (Lin & Huang, 2008), person’s SE, which reflects a confidence in their abilities to use certain systems that is influenced by the actual fit/match of specific technology (D’Ambra et al., 2013; Eom & Estelami, 2012; Tripathi & Jigeesh, 2015), and PBC (Escobar-Rodriguez & Monge-Lozano, 2012; Wu & Chen, 2017). However, no previous studies have examined the relationship between TTF and PBC. Consequently, our third set of hypotheses are:

H7:

Students’ ATT is significantly affected by TTF.

H8:

Students’ SE of using online services is significantly affected by TTF.

H9:

Students’ PBC of using online services are significantly affected by the TTF.

3.2. The impact of personal factors on BI

3.2.1. Self-efficacy (SE)

Several studies have reported the positive correlations between the SE and ATT (Verkijika & De Wet, 2018), the PBC (Hung et al., 2006, 2009; Lee-Partridge & Ho, 2003; Suki & Ramayah, 2010; Susanto & Goodwin, 2011), and BI in the context of e-government (Bwalya, 2012; Terry & O’Leary, 1995). Accordingly, our fourth set of hypotheses are:

H10:

Students’ SE has a significant positive influence on their ATT of online services.

H11:

Students’ SE has a significant positive influence on their PBC of online services.

H12:

Students’ SE has a significant positive impact on BI of using online services.

3.2.2. Attitude (ATT), Perceived Behavioural Control (PBC), and Behavioural Intention (BI)

Individual’s beliefs which enable specific behaviours to be performed are called control beliefs (Zahid & Din, 2019). Numerous studies concur with the opinion expressed by (Chang et al., 2020; Mou & Benyoucef, 2021; Terry & O’Leary, 1995) who consider users’ ATT and PBC to have a positive effect on the BI. Consequently, our fifth set of hypotheses are:

H13:

Students’ ATT has a significant positive influence on their BI of using online services.

H14:

Students’ PBC has a significant positive influence on their BI of using online services.

4. Methodology

4.1. Instrument development and validity

Model items (See Appendix) used in this study were adapted from peer-reviewed literature with some phrasing changes to suit the users and the environmental context (see Table 1). The authors cross checked the ambiguity, understandability, and appropriateness of the items of the questionnaire with six academics. As a result, the researchers used the “expert review” approach, in which they asked six academic experts to examine the questionnaire items and offer input on their clarity, appropriateness, and relevance to the study constructs under consideration. Based on the expert comments, the researchers revised the questionnaire to ensure that it is appropriate for the currently investigated research gap. The English version of the questionnaire was translated into the Arabic language (the mother tongue of Oman) to avoid any confusion or misunderstanding. To ensure the accuracy, validity, and consistency of the English and Arabic versions of the questionnaire a combination of back-translation and expert review (to identify and correct any potential errors or inconsistencies in the translation) methods were used.

Table 1. Instrument scales (items and sources)

Scale	Items	Adapted from
SQ	4	Alruwaie et al. (2020); Klopping and Mckinney (2004); Fathema et al. (2015)
SE	3	Alruwaie et al. (2020), Verkijika and De Wet (2018); Lu et al., (2014)
FC	3	Susanto and Goodwin (2011), Teo (2010); Suki and Ramayah (2010), Hung et al. (2009), Hung et al. (2006)
PBC	3	Lu et al. (2014); Lau (2004), and Terry and O’Leary (1995)
ATT	4	Park et al. (2012), Lu et al. (2014), Lau (2004), and Terry and O’Leary (1995); and Ajzen (1991)
BI	3	Klopping and Mckinney (2004); Park et al. (2012); Fathema et al. (2015)
TTF	7	Goodhue and Thompson (1995); Klopping and Mckinney (2004)

The questionnaire was piloted by 20 students where few modifications were made to the wording. All final structured items were measured using a five-point Likert-type scale, ranking “‘1 = strongly disagree’” and ‘‘5 = strongly agree”. The five-point Likert-type scale was employed in this study because it is easy to administer and assess, offers enough answer possibilities to capture complex replies without overwhelming the respondent, and has been extensively used in prior research, making it easier to compare results across studies.

4.2. Sample

The questionnaire was distributed by the support of relevant directories in the Ministry of Higher Education to all Omani students who use the E-government portal when requesting educational services in Oman. The final paper-based survey was distributed targeting all Omani students across the country with a population of some 55,000 students. The academic portal (students) provides two types of services: 1. Essential Services, for example: completing registration of external scholarship students, completing registration for internal scholarships and grants, completing post graduate registration procedures login page, academic services portal, international co-operation student registration, and request for password. 2 Extra Services, for examples: Oman universities and majors, abroad universities documents, recognition letter for higher educational institutions inside Oman, visa requirements by countries, rules and regulations for Omani students, academic eServices user manuals, equivalency department forms and user manuals, and request for training in MoHERI for students.

A convenient and non-probability sampling method was adopted to recruit the participants, and all were designated based on their availability. All respondents were assured that their confidentiality and anonymity was protected through a cover letter, to comply with ethical research guidelines. In total the number of received responses was 809, only 81% (657) of which were valid for analysis (see Table 2 the demographic profile for the study sample). In the field of IS, the ”10-times rule” strategy is the most often used minimum sample size estimate method in PLS-SEM (Hair et al., 2011; Kock & Hadaya, 2018; Peng & Lai, 2012). The most common variant of this method requires that the sample size be greater than 10 times the maximum number of inner or outer model linkages pointing to any latent variable in the model (Goodhue et al., 2012; Kock & Hadaya, 2018) which is achieved in the current study. Furthermore, given the length of the survey, this response rate was expected and comparable to earlier studies conducted in the same setting (Alraja, 2022; Alraja et al., 2022, 2023; Imran et al., 2022). About 58.75 % of the respondents were female. Most of sample members (64%) held bachelor’s degrees, 30% held a diploma degree and 5.6% held master’s degrees. Most of the citizens were aged between 20 and 30, and 32% were between 30 and 40.

Table 2. Sample characterisation

	Indicator	Frequency	Percent
Gender	male	223	41.25
	female	386	58.75
	Total	657	100%
Education	High diploma	197	30.0
	BA	422	64.2
	Master	37	5.6
	PhD	1	.2
	Total	657	100.0
Age	18 to less than 30	467	61.1
	30 to less than 40	138	32.0
	40 to less than 50	40	5.1
	50 and more	12	1.8
	Total	657	100.0

4.3. Common method bias

The survey of this study adopted a single sources technique in line with Podsakoff et al. (2003) following suggestions that when collecting private data this might lead to circumventions (CMV) of respondents’ willingness to participate in answering freely. Participants could feel that the researcher cannot guarantee their anonymity and confidentiality, i.e. improve response rates, as well as complying with anonymity and confidentiality. Nevertheless, our findings may not be free from the effects of common method bias, because this study measured all variables from the same sources. Accordingly, to reduce CMV, a set of precautions was used both during the questionnaire design and administration, and after collecting our data. As mentioned before to judge the validity, reliability and consistency of our first instrument draft, we used Ping (2004) guidance in our questionnaire design. First, a set of experts (including six academics) were consulted in terms of the questionnaire’s structure and content. The first draft was then modified and was followed by a pilot study using the purposive sampling method of 20 students, which helps to ensure the validity and reliability of our questionnaire before larger scale distribution (Van Teijlingen & Hundley, 2002). The third version was then reviewed considering the collected feedback, resulting in a final instrument. Additionally, all the contributors were briefed on the research aim, with confirmation on the anonymity and confidentiality of their data, while encouraging all respondents to answer the questions independently and truthfully. The adopted constructs were separated randomly in the final distributed questionnaire. After data collection, the Harman’s single-factor test was conducted to verify the presence of CMB. The test showed that there were 6 factors with highest variance accounted for; the first rotated factor was 34.033% (which is less than 50%) (Podsakoff et al., 2012), indicating that the CMB was not a major concern in our study (Pinzone et al., 2019).

5. Data analysis and results

5.1. Measurement validation

The properties of all the variables adopted in the study model are presented in Table 3, which shows that the cross-loading values of all items exceed 0.5 (AL-Alawi, 2017). Providing that the corresponding p-value is significant, this means that each adopted construct is valid and its indicators are internally consistent. The average variance extracted (AVE) for each construct exceeded the recommended threshold of 0.5 (Hair et al., 2016). To demonstrate internal consistency, the study employed Cronbach’s alpha and composite reliability measure. According to Kock and Verville (2012), for Cronbach’s alpha and composite reliability for each construct to be deemed acceptable, sufficient or excellent, the respective values should be 0.7, 0.8, or 0.9 or greater. As seen in Table 3, all the constructs exceeded the acceptable threshold. The above-mentioned results adequately demonstrated convergent validity and internal consistency. The result of the statistical tests for skewness and kurtosis for each construct was between + 2 and − 2 (Alraja et al., 2020). Thus, the responses relating to the study’s constructs were normally distributed.

Table 3. Measurement properties of reflective construct

Construct	Indicator	Mean	SD	Item loading	Cronbach’s alpha	CR	AVE	Skewness	kurtosis
SQ	SQ1	3.51	1.03	0.74	0.79	0.87	0.62	−.348	−.246
	SQ2	3.17	1.16	0.82
	SQ3	3.47	0.98	0.81
	SQ4	3.45	0.98	0.78
SE	SE1	3.58	0.93	0.83	0.77	0.87	0.69	−.456	.199
	SE2	3.56	0.99	0.83
	SE3	3.65	1.01	0.82
FC	FC1	3.54	0.96	0.86	0.75	0.86	0.67	−.488	.384
	FC2	3.52	1.02	0.82
	FC3	3.64	0.92	0.77
TTF	TTF1	3.50	0.92	0.70	0.86	0.89	0.55	−.659	.596
	TTF2	3.52	0.99	0.78
	TTF3	3.57	0.92	0.79
	TTF4	3.53	0.97	0.73
	TTF5	3.57	0.94	0.73
	TTF6	3.68	0.93	0.72
	TTF7	3.56	1.02	0.72
PBC	PBC1	3.90	0.73	0.83	0.70	0.83	0.63	−1.039	1.94
	PBC2	3.93	0.76	0.71
	PBC3	3.95	0.72	0.84
ATT	ATT1	3.66	0.95	0.77	0.80	0.87	0.62	−.654	.650
	ATT2	3.66	0.98	0.81
	ATT3	3.63	0.92	0.79
	ATT4	3.61	0.95	0.78
BI	BI1	3.82	0.84	0.83	0.79	0.88	0.71	−.580	1.006
	BI2	3.76	0.87	0.87
	BI3	3.70	0.94	0.82

The square roots of the AVEs (Fornell-Larcker criterion) were used to identify whether the measurement scale had discriminant validity (Kock & Verville, 2012). The square root of AVE is shown on the diagonal of the correlation matrix in Table 4. The measurement model has adequate discriminant validity if the square roots of the AVEs are larger than any of the correlations involving that construct (Kock, 2015), as well as the value for each construct being at least 0.5 (Fornell & Larcker, 1981).

Table 4. Fornell-Larcker criterion, HTMT, and VIFs

Fornell-Larcker criterion
Variables	SQ	SE	FC	TTF	PBC	ATT	BI
SQ	0.79
SE	0.59	0.83
FC	0.53	0.55	0.82
TTF	0.53	0.49	0.53	0.74
PBC	0.27	0.31	0.29	0.28	0.79
ATT	0.53	0.51	0.49	0.63	0.29	0.79
BI	0.35	0.40	0.38	0.45	0.26	0.48	0.84
HTMT
SQ
SE	0.75
FC	0.68	0.72
TTF	0.63	0.60	0.65
PBC	0.37	0.42	0.39	0.36
ATT	0.66	0.65	0.63	0.76	0.38
BI	0.44	0.51	0.50	0.54	0.35	0.60
VIFs
SQ		1.57			1.81	1.81
SE					1.79	1.79
FC		1.57			1.70	1.70
TTF		1.57			1.61	1.61
PBC							1.10
ATT							1.09
BI

The results of the HTMT test (Table 4) indicated the adequate validity of the measurement model, as all ratios were lower than the most restrictive threshold of 0.85 (Henseler et al., 2015). Moreover, all the values of full collinearity VIFs for this study were lower than 3.3, which strongly supports the absence of multicollinearity in the adopted model and no common method bias.

5.2. Estimation model and its robustness

Based on the results generated by the WarpPLS software 7.0, which provides model fits and quality indices (Table 5), quality indices were established for all the criteria used in this study (Kock, 2019a). The robustness of the model was tested by implementing the following steps. First, as suggested by Kock (2019a), the structural model was run using the (bootstrap, jack-knifing and stable 4) procedure of resampling. The p-values were compared across these methods, showing the same stability. Second, to assess the predictive validity associated with each latent variable block in the model, Stone-Geisser Q-squared coefficients were used (Geisser, 1974; Stone, 1974). The results displayed in Table 6 refer to acceptable predictive validity, as all Q-squared coefficients were greater than zero. Third, in relation to average communality, the goodness of fit (0.653) was conducted along with the quality of the complete structural model in terms of average R2 coefficients, which are conducted only for endogenous latent variables. Also, all R-squared coefficients exceeded the general recommendations made by Cohen (1988). Values of R-squared coefficients below 0.02 were considered for revision, as the explanatory power in sub-models was below reasonable expectations (Kock, 2019b).

Table 5. Model fit and quality indices

Criterion	Assessment	Acceptable if	Supports Model fit and quality indices
Average Path Coefficient (APC)	APC = 0.204, P<0.001	P≤0.05	yes
Average R-squared (ARS)	ARS=0.324, P<0.001	P ≤ 0.05	yes
Average Adjusted R-squared (AARS)	AARS=0.321, P<0.001	P ≤ 0.05	yes
Average Block VIF (AVIF)	AVIF =1.595	AVIF ≤ 5, ideally ≤ 3.3	yes
Average Full Collinearity VIF (AFVIF)	AFVIF =1.713	AFVIF≤ 5, ideally ≤ 3.3	yes
Tenenhaus GoF (GoF)	GoF =0.455	small ≥ 0.1, medium ≥ 0.25, large ≥ 0.36	yes
Sympson’s Paradox Ratio (SPR)	SPR =1.000	SPR ≥ 0.7, ideally = 1	yes
R-squared Contribution Ratio (RSCR)	RSCR=1.000	RSCR≥ 0.9, ideally = 1	yes
Statistical Suppression Ratio (SSR)	SSR=1.000	SSR ≥ 0.7	yes
Nonlinear Bivariate Causality Direction Ratio (NLBCDR)	NLBCDR=1.000	NLBCDR ≥ 0.7	yes
Standardized Root Mean Squared Residual (SRMR)	SRMR=0.065	acceptable if ≤ 0.1	yes
Standardized Mean Absolute Residual (SMAR)	SMAR=0.050	acceptable if ≤ 0.1	yes
Standardized Chi-squared (SChS)	(SChS)=6.869, P<0.001	P ≤ 0.05	yes

Table 6. Model robustness

Construct	R^2	Communality (AVE)	Q^2
SQ	–	0.62	–
SE	0.443	0.69	0.444
FC	–	0.67	–
TTF	–	0.55	–
PBC	0.131	0.63	0.133
ATT	0.482	0.62	0.483
BI	0.270	0.71	0.272
Average	0.32	^a0.64	0.33

^aThe average of communality is computed as a weighted average of all of the communalities using weights as the number of manifest variables in each construct with at least two manifest indicators.

5.3. Hypotheses test

PLS-SEM was adopted for hypotheses testing (Mensah et al., 2021) as shown in Figure 2 and Table 7. Following the recommendations of Kock (2014), different testing methods were used to investigate the mediation effect (Preacher, 2015), by checking the following: the direct effect of independent variables on the dependent variable with and without including the mediation role; the direct effect of each independent variable on the mediator variable and coded (a); the effect of the mediator variable on the dependent variable and coded (b); the Standard Error of path (a); and the Standard Error of path (b).

Figure 2. Tested model.

Table 7. PLS-SEM model

Path (Direct Effects)	Coefficient	P value	T-test	Hypothesis	Results
SQ → SE	0.364	<0.001	9.701	H1	accepted
SQ → PBC	0.082	0.017	2.126	H2	accepted
SQ → ATT	0.161	<0.001	4.194	H3	accepted
SE → PBC	0.145	<0.001	3.774	H4	accepted
SE → BI	0.203	<0.001	5.319	H5	accepted
SE → ATT	0.185	<0.001	4.827	H6	accepted
FC → SE	0.279	0.002	7.357	H7	accepted
FC → PBC	0.095	0.007	2.464	H8	accepted
FC → ATT	0.080	0.019	2.080	H9	accepted
TTF → SE	0.147	<0.001	3.823	H10	accepted
TTF → PBC	0.128	<0.001	3.331	H11	accepted
TTF → ATT	0.414	<0.001	11.089	H12	accepted
PBC → BI	0.132	<0.001	3.421	H13	accepted
ATT → BI	0.434	<0.001	11.656	H14	accepted

Based on the aforementioned results, Sobel’s test was executed to check the significance of the mediation, coefficient of the mediator, T-test of mediating effect and p-value. All the results of the above-mentioned tests confirmed the results, as presented in Table 7.

6. Discussion

6.1. Theoretical implications

Our study has examined the effect of system quality, facilitating conditions and technology task fit upon students’ perceptions of online services. The results highlight the importance of system quality when considering the effect on self-efficacy. These results corroborate the findings of previous work, such as Alruwaie et al. (2020). The system quality does show some significant effect on the attitude to use the online ministry services, compared to a lower effect on the perceived behaviour control. Hence, it could plausibly be hypothesised that students’ confidence using a system’s features, functions and content, is highly related to the system’s quality, i.e. system interaction, functions, system content and internet speed. In addition, these results reveal that the self-efficacy is directly affected by the facilitating conditions, which is in accordance with the findings of Ching-Ter et al. (2017); Hung et al (2009, 2006);. Lemay et al. (2018); Mahat et al. (2012); Sánchez-Prieto et al. (2017); Suki and Ramayah (2010); Susanto and Goodwin (2011). Nevertheless, the facilitating conditions were found to have some positive effects on the attitude and perceived behaviour control, which is consistent with the findings of Abdullah and Toycan (2017); García Botero et al. (2018); Teo (2010). This means that the accessibility and availability of the important resources (e.g. portal guidance, online assistance or specialized instructions) on how to use the online services, has more effect on the students’ perceptions and confidence in their ability to use the portal’s features, functions and online content, compared to the students’ attitudes and their perceived behaviour control. The results further support the idea of considering the technology task fit as being the key dimension to influencing students’ attitudes towards using the online services. Another important finding was that a direct relationship exists between self-efficacy and the perceived behaviour control. This means that as students’ confidence and belief in their ability to use the online government services increases, the easier it becomes to use these services. Consequently, students’ confidence in and judgement of their skills to use these services is a significant variable in increasing both students’ behavioural intention and the perceived behaviour control of the online services. These results are consistent with the conclusions of Bwalya (2012); Hung et al (2009, 2006).

A positive correlation was found between technology task fit as an independent factor with the dependent factors according to the relation strength for attitude (Lin & Huang, 2008), self-efficacy (D’Ambra et al., 2013; Eom & Estelami, 2012; Tripathi & Jigeesh, 2015), perceived behaviour control (Escobar-Rodriguez & Monge-Lozano, 2012; Wu & Chen, 2017). These findings suggest that the information on the online ministry services should be clearly and sufficiently detailed, quick and easy to access, accurate and current, and lastly, easily stored and retrieved when students need to use it. If the information in these services has those attributes, students’ attitudes will be the personal factor which most affects their leaning towards using the online services; therefore, they are more likely to use it and perform tasks more efficiently.

Our results indicate that technology task fit is a strong technological factor of theory of planned behaviour, affecting all personal factors, especially the attitude. Therefore, the higher the technology task fit of portals, the more likely the positive influence of students’ attitudes and accordingly, the intention to use the online services. By connecting the portal functions to perform specific tasks (e.g. modify personal information), this allows students to both positively change their attitude, and have continued intentions of using it. This result is in harmony with the findings of Lin (2012), Lin and Huang (2008), signifying that technology task fit had a positive effect on attitude and the behavioural intention. Thus, as technology task fit increases, students are more likely to change their attitudes and intentions to use it. Moreover, the larger the technology task fit, the higher the chance that the online ministry services will be used with more self-confidence and will be more behaviour controlled.

Adding self-efficacy to theory of planned behaviour factors has been found to contribute strongly to the theory. The technological factors (technology task fit, facilitating conditions and system quality) together, count as 44.3% of the effect on the SE in the model, which in turn affects the behavioural intention. The implication is that by providing accurate information, important functions/content, and the necessary guidance and assistance to achieve tasks more effectively, the students’ confidence in using the online ministry services will continue to increase and promote their intentions to use the services in future. System quality was the most correlated factor to the SE among all the technological factors. This outcome shows that system quality had a strong impact on the students’ confidence for prolonged use of the online services. Facilitating conditions was the second to have a significant positive influence on the self-efficacy of these services. Therefore, managers should pay close attention to this when developing the facilitating conditions.

We also found that merging technology task fit, system quality, facilitating conditions, and self-efficacy variables relating to behavioural intention produces a stronger and new association of the variables that relate to the theory of planned behaviour factors (attitude and perceived behaviour control) than adopting them separately. For example, self-efficacy affects behavioural intention compared to the effect of perceived behaviour control on behavioural intention. Also, the three new added technological factors (facilitating conditions, technology task fit and system quality) together, count as 48.2% of the attitude effect in theory of planned behaviour. Despite the presence of a direct positive relationship between perceived behaviour control and the behavioural intention, among the effects of the three technological factors (self-efficacy, technology task fit and facilitating conditions), technology task fit has the most positive effect of the perceived behaviour control. This new result in our research contributes to the theory of planned behaviour as original findings. Thus, the perceived ease or difficulty in performing, the behaviour to use the online services should take into account having sufficient easy, accurate and detailed information. Our research is one of the few studies that have contributed to theory of planned behaviour by adding multiple factors towards using the online government services by students, in order to enrich understanding of the online services behaviour intention.

6.2. Practical implications

The overall implication for educational administration is to be aware of the real needs of the students when providing an e-service. Technically, the study highlights the importance on how to increase the portal’s quality, task fitness and provide important conditions when planning to inspire the students to use the system. In other words, designers would need to: include inbuilt training on how to use the online government service functions effectively; offer guidance and adequate details to users on how to get the content and information quickly and easily from the portal; design interactive functions that can locate the information easily and display it in easy to read and understandable formats; and encourage the user to develop confidence about using online content, features and functions.

7. Limitations and future research

This study has some limitations that could be addressed in future research. Firstly, the model developed as part of this research focused on the TTF only; other constructs such as individual technology fit and social factors can be investigated directly or indirectly to understand the impact these make on students’ BI. Secondly, future research could take a longitudinal approach, as students’ preferences and perceptions might be different across their study lifecycle. Thirdly, the sample studied was specific to Oman; by expanding the population to include other Gulf countries such as the Emirates and Saudi Arabia, the results may be more generalisable. Fourthly, our study was conducted in a developing country; a comparative study that explores similarities and differences in the attitude of students in developing and developed countries could help developing economies to progress. Finally, in this study, the proposed model has not considered all factors, and other factors such as gender, age, and cultural dimensions could be included for further investigation.

8. Conclusion

Online government portal designers need to consider the students’ intentions; not only their attitude, but also on self-efficacy, to stimulate their intentions to continue using these services. Our findings imply that students’ intentions can be improved by augmenting their awareness and beliefs that the e-services portals can strengthen the effectiveness of their tasks. On the other hand, our study confirms the need to design the online services portal in line with existing students’ requirements to perform specific tasks, including the resources available. The outcome of this research may help other government organisations in the Gulf countries to understand their students’ needs, preferences and perceptions, and provide insights to help design, implement and manage their portal services.

Our study differs from other research in a few ways as follows: it developed a new combined model in the context of the Omani e-government; it compared other studies in the Omani context, such as Al-Busaidy and Weerakkody (2009), Al-Mamari et al. (2013) and Sharma et al. (2017), by providing a robust recent literature review relating to the adoption of online government services; and it is one of the first attempts at thoroughly examining the direct impact of technology task fit on perceived behaviour control, and investigating in more detail the correlations between facilitating conditions and self-efficacy in the e-government context which differ from other studies that explore the direct facilitating conditions and technology task fit effect on behavioural intention.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The work was supported by the The Research Council (TRC), Sultanate of Oman (Block Fund-Research Grant), BFP/RGP/ICT/21/132.

Appendix

System Quality (SQ)

SQ1: I am satisfied with the academic e-service portal functions

SQ2: I am satisfied with the Internet speed when using the academic e-service portal

SQ3: I am satisfied with the academic e-service portal content

SQ4: I am satisfied with the academic e-service portal interaction

Self-Efficacy (SE)

PSE1: I feel confident using academic e-service portal features

PSE2: I feel confident operating academic e-service portal functions

PSE3: I feel confident using Online learning content in an academic e-service portal

Facilitating conditions (FC)

FC1: When I need help using the academic e-service portal guidance is available to me

FC2: A specific person/group is available for assistance with any difficulties related with academic e-service portal use

FC3: Specialized instruction concerning academic e-service portal use is available to me

Perceived Behavioural Control (PBC)

PBC1: I have enough opportunities to use the academic e-service portal.

PBC2: I have the capacity to use the academic e-service portal

PBC3: I have enough control over using the academic e-service portal

Attitude (Att)

ATT1: I think it is worthwhile to use the academic e-service portal

ATT2: I like using the academic e-service portal

ATT3: In my opinion, it is very desirable to use an academic e-service portal for academic and related purposes

ATT4: I have a generally favourable attitude toward using the academic e-service portal

Behavioural Intention to Use (BI)

BI1: I intend to use the functions and content of the academic e-service portal to assist my academic activities

BI2: I intend to use the functions and content of the academic e-service portal as often as possible

BI3: I intend to use the functions and content of the academic e-service portal in the future

Task-Technology Fit (TTF)

TTF1: Sufficiently detailed information is maintained on the academic e-service portal website

TTF2: On the academic e-service portal, detailed information is either obvious or easy to find out

TTF3: I can get information quickly and easily from the academic e-service portal when I need it

TTF4: The online academic e-service portal information that I use or would like to use is accurate enough for my purposes

TTF5: The online academic e-service portal information is up to date enough for my purposes

TTF6: The online academic e-service portal information that I need is displayed in a readable and understandable form

TTF7: The online academic e-service portal information maintained at academic e-service portals pretty much what I need to carry out my tasks